Descriptive and Analytic Studies
Presenter’s Name Presenter’s Title
Title of Event Date of Event
Department of Health and Human Services Centers for Disease Control and Prevention Learning Objectives
• Identify the following for an NCD problem: • Type of study to conduct • Sampling methods to use • Measure of association to calculate for a particular study • Interpret the results of descriptive and analytic studies.
Descriptive and Analytic Studies 2 Lesson Overview
• Reasons for conducting studies • Definition, characteristics, and analysis of: • Descriptive studies • Analytic studies • Methods of sampling
Descriptive and Analytic Studies 3 Why Conduct Studies?
To describe burden of disease or prevalence of risk factors, health behaviors, or other characteristics of a population that influences risk of disease • To determine causes or risk factors for illness • To determine relative effectiveness of interventions
Descriptive and Analytic Studies 4 Taxonomy of Epidemiologic Studies: Figure 1
Descriptive and Analytic Studies 5 Descriptive or Analytic Studies?
Descriptive studies • Generate hypotheses • Answer what, who, where, and when
Analytic studies • Test hypotheses • Answer why and how
Descriptive and Analytic Studies 6 DEFINITION AND CHARACTERISTICS OF DESCRIPTIVE STUDIES Descriptive and Analytic Studies 7 Descriptive Studies
Characterize who, where, or when in relation to what (outcome) • Person: characteristics (age, sex, occupation) of the individuals affected by the outcome • Place: geography (residence, work, hospital) of the affected individuals • Time: when events (diagnosis, reporting; testing) occurred
Descriptive and Analytic Studies 8 Types of Descriptive Studies
Aggregate Individual
Ecological Case Report Studies
Case Series
Cross- sectional Study
Descriptive and Analytic Studies 9 Cross-Sectional Study as a Descriptive Study Purpose: To learn about the characteristics of a population at one point in time (like a photo “snap shot”)
Design: No comparison group
Population: All members of a small, defined group or a sample from a large group
Results: Produces estimates of the prevalence of the population characteristic of interest
Descriptive and Analytic Studies 10 When to Conduct a Cross- Sectional Study • To estimate prevalence of a health condition or prevalence of a behavior, risk factor, or potential for disease
• To learn about characteristics such as knowledge, attitude and practices of individuals in a population
• To monitor trends over time with serial cross- sectional studies
Descriptive and Analytic Studies 11 Cross-Sectional Study Measures
Prevalence of a condition:
= number of existing cases / size of population
(or population count)
Descriptive and Analytic Studies 12 Example: Cross-Sectional Study Objective • To estimate the magnitude and patterns of violence against pregnant women Study • Population-based, household, cross-sectional study in Mbeya and Dar es Salaam, Tanzania, 2001-2002 Result • Violence experienced by 7% in Dar es Salaam and 12% in Mbeya
Ref: Stöckl H, Watts C, Kilonzo Mbwambo JK. Physical violence by a partner during pregnancy in Tanzania: prevalence and risk factors. Reprod Health Matters. 2010 Nov;18(36):171-80.
Descriptive and Analytic Studies 13 Studies to Track Trends in Newly Recognized Cases Incidence study • Newly reported or registered disease cases compared over time, place, or person • Population estimates or other population group totals used as denominators
Ecological study • Rates are linked to the level of exposure to some agent for the group as a whole
Descriptive and Analytic Studies 14 Example: Incidence Study Objective • To estimate the incidence and prevalence of diabetes in young persons in the United States Study • Annual diabetes death rates among youth aged <19 calculated from National Vital Statistics System data from 1968-2009 Result • Trends for diabetes death rates varied by age group
Saydah, S, Imperatore, G., Geiss, L., & Gregg, E. (2012). Diabetes death rates among youths aged <19 years—United States, 1968-2009. MMWR, 61(43), 869-871
Descriptive and Analytic Studies 15 Example Incidence Study (continued)
Saydah, S, Imperatore, G., Geiss, L., & Gregg, E. (2012). Diabetes death rates among youths aged <19 years—United States, 1968-2009. MMWR, 61(43), 869-871.
Descriptive and Analytic Studies 16 Taxonomy of Epidemiologic Studies: Figure 2
Descriptive and Analytic Studies 17 Analytic Studies Definition
Analytic studies test hypotheses about exposure- outcome relationships
• Measure the association between exposure and outcome
• Include a comparison group
Descriptive and Analytic Studies 18 Developing Hypotheses
• A hypothesis is an educated guess about an association that is testable in a scientific investigation. • Descriptive data (Who? What? Where? When?) provide information to develop hypotheses. • Hypotheses tend to be broad initially and are then refined to have a narrower focus.
Descriptive and Analytic Studies 19 Developing Hypotheses Example Hypothesis: People who smoke shisha are more likely to get lung cancer than people who do not smoke shisha. • Exposure: smoking shisha • Outcome: lung cancer
Hypothesis: ? • Exposure: ? • Outcome: ?
Descriptive and Analytic Studies 20 Analytic Study Types
Experimental Observational Studies Studies
Randomized Cohort Control (Intervention) Trials
Case-control
Cross-sectional
Descriptive and Analytic Studies 21 Cohort Studies
What is a cohort?
A well-defined group of individuals who share a common characteristic or experience
• Example: Individuals born in the same year
What are other examples of cohorts?
Descriptive and Analytic Studies 22 Cohort Study (longitudinal study, follow-up study) • Participants classified according to exposure status and followed-up over time to ascertain outcome • Can be used to find multiple outcomes from a single exposure • Appropriate for rare exposures or defined cohorts • Ensures temporality (exposure occurs before observed outcome)
Descriptive and Analytic Studies 23 Cohort Study Design
Disease Exposed No Disease
Study Exposure is Follow Population self-selected over time Disease Unexposed No Disease
Descriptive and Analytic Studies 24 Types of Cohort Studies
Prospective cohort studies • Group participants according to past or current exposure and follow-up into the future to determine if outcome occurs
Retrospective cohort studies • At the time that the study is conducted, potential exposure and outcomes have already occurred in the past
Descriptive and Analytic Studies 25 Prospective Cohort Studies Disease Exposed No Disease Study Population Disease Unexposed No Disease
Start of study (Present) (Future) Descriptive and Analytic Studies 26 Retrospective Cohort Studies
Disease Exposed No Disease Study Population Disease Unexposed No Disease
Start of study (Past) (Present)
Descriptive and Analytic Studies 27 When to Conduct a Cohort Study When the exposure is rare and the outcome is common • Agricultural pesticide use and cancer events
To learn about multiple outcomes due to a single exposure • Health effects of a nuclear power plant accident
Descriptive and Analytic Studies 28 Analysis of Cohort Studies
Risk: Quantifies probability of experiencing the outcome of interest in a given population • Calculation: Number of new occurrences of outcome/population at risk Example: • 29 new cases of diabetes in a community • 100,000 people in the community at risk for diabetes • What is the risk of diabetes? 29/100,000
Descriptive and Analytic Studies 29 Analysis of Cohort Studies: Person-Time, Rate Quantifies occurrence of outcome in population by time Calculation: number of new cases during follow-up period Sum of time each study participant was followed and at risk of disease Example: 1,212 tunnel workers 160 deaths among tunnel workers 24,035 person-years at risk Mortality rate = 160 / 24,035 = 6.7 deaths per 1,000 workers per year
Ref:. Stern et al. Heart Disease Mortality Among Bridge and Tunnel Officers Exposed to Carbon Monoxide. American Journal of Epidemiology.1988;128:1276-1288
Descriptive and Analytic Studies 30 Risk Ratio
• Can also be called Relative Risk or RR • Quantifies a population’s risk of disease from a particular exposure • Calculation: Risk in the exposed group / Risk in the unexposed group
Descriptive and Analytic Studies 31 Rate Ratio
Compares the rates of disease in two groups that differ by demographic characteristics or exposure history
Calculation: Rate for group of primary interest Rate for comparison group
Descriptive and Analytic Studies 32 RR Strength Scales
RR Strength RR 0.71 – 0.99 Weak 1.01 – 1.50 0.41 – 0.70 Moderate 1.51 – 3.00 0.00 – 0.40 Very strong >3.00
Oleckno WA. Essential epidemiology: principles and applications. Prospect Heights, IL 2002;108.
Descriptive and Analytic Studies 33 Example: Risk Ratio
Question: What is the relationship between being obese and getting type 2 diabetes?
Risk in the exposed group (obese) = .00076 = 5.8 Risk in the unexposed group (non-obese) .00013
Risk Ratio = 5.8
Interpretation: The risk of diabetes among those who are obese is 5.8 times the risk among those who are not obese.
Descriptive and Analytic Studies 34 Example: Person-Time Rate Ratio NHANES – Follow-up Study (male diabetics subset) • Original enrollment 1971- 1975 • Follow-up 1982 – 1984 • Complete follow-up on:
Enrolled Died PY of F/U Diabetics 189 100 1414.7 Non-diabetics 3151 811 28,029.8
• Mortality Rate Ratio: • 100/1414.7 ÷ 811/28,029.8 = 70.7/1000 ÷ 28.9/1000= 2.5
Ref: Kleinman J, et al. Am J Epidemiol. 1988; 128:389-401.
Descriptive and Analytic Studies 35 Case-Control Study
Purpose: • To study rare diseases • To study multiple exposures that may be related to a single outcome
Study Subjects Participants selected based on outcome status: • Case-subjects have outcome of interest • Control-subjects do not have outcome of interest
Descriptive and Analytic Studies 36 Case-Control Study Design
Exposed Cases (Diseased)
Unexposed Identify Assess cases Source exposure and Population history select controls Exposed Controls Unexposed (No Disease)
Descriptive and Analytic Studies 37 When to Conduct a Case-Control Study
• The outcome of interest is rare
• Multiple exposures may be associated with a single outcome
• Funding or time is limited
Descriptive and Analytic Studies 38 Case-Control Study: Analysis Format Exposure Cases Controls
Yes a b
No c d
Exposure odds ratio (OR) ≈ RR when disease is rare
Odds of being exposed among the cases = a/c Odds of being exposed among the controls = b/d
Exposure odds ratio = (a/c)/(b/d) = (a*d)/(b*c) (Cross-product ratio)
Descriptive and Analytic Studies 39 Example Odds Ratio Lead Poisoning
Work in mine? Cases Controls Yes 17 13 No 83 87
Odds Ratio = 17/83 ÷ 13/87 = 17x87 / 13x83= 1.37
Descriptive and Analytic Studies 40 Prevalence Ratio and Prevalence Odds Ratio • Chronic disease – date of onset is unknown
• Measure prevalence rather than incidence
RR PR (prevalence ratio) OR POR (prevalence odds ratio)
Descriptive and Analytic Studies 41 Prevalence Ratio
• Usually from a cross-sectional study • Similar to risk ratio from cohort study
Exposure With disease Without disease Total Exposed a b a+b Unexposed c d c+d Total a+c b+d
• PR= Prevalence of disease in exposed group/ Prevalence of disease in unexposed group OR • PR= a/(a+b) / c/(c+d)
Descriptive and Analytic Studies 42
Prevalence Odds Ratio
• Usually from a cross-sectional study • Similar to odds ratio from case control study • Calculated same way as odds ratio: POR = a*d c*b
With disease Without disease Exposed a b a+b
Unexposed c d c+d a+c b+d
Descriptive and Analytic Studies 43 Example: Prevalence Ratio and Prevalence Odds Ratio Prevalence of Breast Cysts Lifetime use of oral Yes No Total contraceptives Cyst Cyst Ever Used 124 3123 3247 Never Used 77 2557 2644 Total 201 5690 5891 Prevalence of breast cysts among ever users = 124/3247 = .038 Prevalence of breast cysts among never-users = 77/2644 = .029 Prevalence ratio = .038/.029 = 1.3 Prevalence odds ratio = 124 * 2557 3123 * 77 = 1.3
Descriptive and Analytic Studies 44 Practice Exercise #1
Background: • NCDs such as type 2 diabetes are poorly understood and under-prioritized in many low-to-middle income countries. • You want to determine the risk of type 2 diabetes associated with cardiovascular risk factors such as obesity and abdominal fat mass in your country. Questions: 1. What type of study would you conduct and why? 2. What is the measure of association to calculate for this study?
Descriptive and Analytic Studies 45 Practice Exercise #2
Background: • The prevalence of prostate cancer has increased in your country over the last 5 years. • You want to examine the association between calcium intake and prostate cancer risk. • You have limited time and funding to conduct this study. Questions: 1. What type of study would you conduct and why? 2. What is the measure of association to calculate for this study?
Descriptive and Analytic Studies 46 Practice Exercise #3
Background: • Cardiovascular disease (CVD) is of growing concern; however your country has no recent data on the burden of this disease. • You want to estimate the burden of cardiovascular disease in the two main cities in your country. Questions: 1. What type of study would you conduct and why? 2. What is the measure of association to calculate for this study?
Descriptive and Analytic Studies 47 METHODS OF SAMPLING
Descriptive and Analytic Studies 48 Discussion Question
Why do we use sampling?
• Cannot get information on everyone in a population • Efficiently gets information on a large population • Obtains a representative sample of a population
Descriptive and Analytic Studies 49 Sampling Methods
Two main types of sampling methods: • Probability sampling • Non-probability sampling
Descriptive and Analytic Studies 50 Probability Sampling
What are types of probability-based samples?
• Simple random sampling • Systematic random sampling • Stratified random sampling • Cluster sampling
Descriptive and Analytic Studies 51 Simple Random Sample
Principle • Equal chance/probability of drawing each unit
Procedure • List all units (persons) in a population • Assign a number to each unit • Randomly select units
Descriptive and Analytic Studies 52 Method: Simple Random Sampling
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15
16 17 18 19 20 21 22 23 24 25 26 27 28 29 30
Each unit has the same probability of selection (1/30)
Descriptive and Analytic Studies 53 Example: Simple Random Sample Example: Calculate the prevalence of tooth decay among 1200 children attending a school
(sample size =100)
• List all children attending the school • Each child assigned a number from 1 to 1200 • Randomly select 100 numbers between 1 and 1200
Descriptive and Analytic Studies 54 Advantages & Disadvantages: Simple Random Sample Advantages • Simple
Disadvantages • Need complete list of units • Units may be scattered and poorly accessible
Descriptive and Analytic Studies 55 Systematic Random Sample
Principle • Select sample at regular intervals based on sampling fraction Procedure • List all units (persons) in a population • Assign a number to each unit • Calculate sampling fraction (population size ÷ sample size) • Select first unit at random based on sampling fraction • Subsequent units are chosen at equal intervals
Descriptive and Analytic Studies 56 Advantages & Disadvantages: Systematic Random Sample Advantages • Simple • Can be implemented easily without software Disadvantages • Need complete list of units
Descriptive and Analytic Studies 57 Example: Systematic Random Sample Example: Calculate the prevalence of tooth decay among 1200 children attending a school (sample size =100) • List all children attending the school • Randomize the list to avoid bias • Each child assigned a number from 1 to 1200 • Sampling fraction =1200/100 = 12 • Randomly select a number between 1 and 12 • Example: 8 • Select every 12th child, starting with child #8 • Example: 8, 20, 32, 44… Descriptive and Analytic Studies 58 Stratified Random Sample
Principle • Select random samples from within homogeneous subgroups (strata)
Procedure • List all units (persons) in a population • Divide the units into groups (called strata) • Assign a number to each unit within each stratum • Select a random sample from each stratum • Combine the strata samples to form the full sample
Descriptive and Analytic Studies 59 Method: Stratified Random Sample Men Women • Sampling frame divided into groups (age, sex, 1 2 3 4 5 1 2 3 4 5 socioeconomic status)
6 7 8 9 10 • Units in each group 6 7 8 9 10 have the same
probability of selection, 11 12 13 14 15 11 12 13 14 15 but probability differs between groups 16 17 18 19 20
Probability:1/20 Probability: 1/15
Descriptive and Analytic Studies 60 Advantages & Disadvantages: Stratified Random Sample Advantages • Can get separate estimates from the whole population and from individual strata (if sample is large enough) • Precision increased if less variability within strata than between strata
Disadvantages • Can be difficult to identify strata
Descriptive and Analytic Studies 61 Class Discussion Question
What are some examples of strata that you might sample within?
• Race/ethnicity/tribe/nationality • Smoking status • Age group • Occupation • Gender • Education • Geographic location • Many possibilities! • Socioeconomic status
Descriptive and Analytic Studies 62 Example: Stratified Random Sample Example: Calculate the prevalence of tooth decay among 1200 children attending a school, with equal representation of males and females (sample size =100) • List all children attending the school • Divide the children into two groups • 540 males and 660 females • Assign each child a number • Males: 1 to 540 • Females 1 to 660 • Randomly select 50 males and 50 females
Descriptive and Analytic Studies 63 Cluster Sample
Principle • Select all units within randomly selected geographic clusters Procedure • Divide population into geographic groups (clusters) • Assign a number to each cluster • Randomly select clusters • Sample all units within selected clusters OR select a random sample of units within selected clusters
Descriptive and Analytic Studies 64 Advantages & Disadvantages: Cluster Sample Advantages • List of sampling units not required • More efficient for face-to-face interviews when units are dispersed over a large area
Disadvantages • Loss of precision due to correlation within clusters • This correlation needs to be taken into account in sample size calculations and analysis (“design effect”) Descriptive and Analytic Studies 65 Non-probability Sampling
• Probability of selection is unknown or zero • Inexpensive • Results not generalizable • Results often biased Common types of non-probability sampling: • Convenience sampling • Snowball sampling / Respondent-driven sampling • Voluntary sampling
Descriptive and Analytic Studies 66 Choosing a Sampling Method
Consider: • Population to be studied • Size/geographic distribution • Availability of list of units • Heterogeneity with respect to variable • Level of precision required • Resources available
Descriptive and Analytic Studies 67 Practice Exercise #4
Background: You will choose a sampling method for each of the following studies.
Questions: What sampling method would you use for: 1. The cross-sectional study on CVD described in Practice Exercise #3? Why? 2. A one-time survey of citizens’ attitudes toward smoking and second-hand smoke in response to proposed legislation to impose a ban on smoking in restaurants. Why? 3. Serosurvey of blood lead levels (or urinary arsenic levels) of prisoners entering the nation’s largest prison (or pregnant women entering the nation’s largest maternity ward) to determine average level of exposure in the population.
Descriptive and Analytic Studies 68 SUMMARY
Descriptive and Analytic Studies 69 Descriptive vs. Analytic Epidemiology Descriptive epidemiology: • Who, What, When, and Where
Analytic epidemiology: • Why and How
Descriptive and Analytic Studies 70 Types of Descriptive and Analytic Studies Types of descriptive studies • Aggregate: Ecological study • Individual: Case report, case series, cross- sectional study
Types of analytic studies • Experimental: Randomized control trial • Observational: Cohort, case-control, cross- sectional
Descriptive and Analytic Studies 71 Cohort vs. Case-Control Studies
Study Comparison Cohort Study Case-Control Study Preferred Study Members are easily Identifying entire cohort Design When… identifiable would be too costly or time consuming Members are easily accessible Accessing entire cohort would be too costly or time Exposure is rare consuming
There may be multiple Illness is rare diseases involved There may be multiple exposures involved Study Group Exposed persons Persons with illness (cases) Comparison Group Unexposed persons Persons without illness (controls)
Descriptive and Analytic Studies 72 Sampling Advantages and Disadvantages Probability Sampling Non-Probability Sampling
Advantages Advantages • Results are generalizable • Easy • Representative • Quick access to certain groups Disadvantages • Expensive Disadvantages • Logistically difficult • Not representative • Time-intensive • Results are not generalizable
Descriptive and Analytic Studies 73 Skill Assessment
• You will work in small groups to complete two parts of a skill assessment: 1. Identify the type of study to conduct and sampling method 2. Interpret the results • Materials and questions can be found in your Participant Guide. • Spend approximately 1 hour completing the assessment. • Be prepared to share the group’s work.
Descriptive and Analytic Studies 74 Centers for Disease Control and Prevention (CDC). Descriptive and Analytic Studies. Atlanta, Georgia: Centers for Disease Control and Prevention (CDC); 2013.
Descriptive and Analytic Studies 75 For more information please contact Centers for Disease Control and Prevention 1600 Clifton Road NE, Atlanta, GA 30333 Telephone: 1-800-CDC-INFO (232-4636)/TTY: 1-888-232-6348 Visit: www.cdc.gov | Contact CDC at: 1-800-CDC-INFO or www.cdc.gov/info
The findings and conclusions in this report are those of the authors and do not necessarily represent the official position of the Centers for Disease Control and Prevention.
Department of Health and Human Services Centers for Disease Control and Prevention 76